1
Leveraging Analytics to Change Opioid Prescribing Behavior
Session BP3, February 11, 2019
Mark Binstock, MD, MPH; CMIO Bon Secours Mercy Health
2
Mark Binstock, MD, MPH
Has no real or apparent conflicts of interest to report.
Conflict of Interest
3
• Analytics
• Prevention
• Screening
• Treatment
Agenda
4
• Explore the key metrics for opioid prescribing
• Confirm the central role of morphine equivalents in opioid
analytics
• Review key provider facing tools to reduce inappropriate opioid
prescribing
• Emphasize the contributions of People and Process to success
Learning Objectives
5
‘Mercy Health earns HIMSS Davies
Award for innovative approach to
opioid fight.’
“The Ohio health system implemented analytics and
decision support to reduce opioid prescriptions.”
- HealthcareIT News
https://www.healthcareitnews.com/news/mercy-health-earns-himss-davies-award-innovative-approach-
opioid-fight
5
6
CDC National Guidelines: A Foundation for Our
Build and Analytics
6
https://www.cdc.gov/drugoverdose/prescribing/guideline.html
7
Outpatient Opioid Prescribing
Data Summary
-12%
-16%
-30%
-15
-50%
-20%
Significant Reductions in Opioid Prescribing Metrics
The following opioid ordering behaviors were substantially reduced
between December 2018 and December 2018:
Total Opioid
Orders
Opioid Orders to
All Medication
Orders
Opioids Orders
w/ MEDD > 80
Opioids w/ Day
Supply > 7 days
Total MEDD
per pt
Opioids
Orders w/
MEDD >30
Acute Pain
7
8
Morphine Equivalents
The key to opioid analytics:
– Analogy of calories to food
– Method of quantifying any opioid order
(prescription)
– Morphine Equivalent Daily Dose (MEDD):
• Potency (conversion factor) X Dose (milligrams)
X Frequency per day
– Can also be used at an order, patient, ordering
provider, specialty, regional and population level:
• Potency X Dose X Dispense quantity
8
9
Analytics Methodologies - Core
Opioid Metrics
• Total Opioid Orders: Raw count of narcotic orders within reporting period
• Auth Provider Outlier: Indicates that on any of the 3 Sentinel metrics the
provider scores BELOW the 10th percentile
• Avg Percentile for 3 Sentinel Metrics: Mean of the Percentiles of the 3
Sentinel metrics
• Total MEQ RX: Cumulative morphine equivalent burden for all narcotics
ordered within reporting period. gold standard takes into consideration
potency, dose, frequency, and quantity dispensed
• Percentile Auth Provider Total MEQ RX: This represents the percentile score
of the provider compared with peers with lower percentiles (in red shades)
being associated with higher MEQ and higher percentiles (in green shades)
associated with lower MEQ
10
Analytics Methodologies - Core
Opioid Metrics
• Rate MEQD GT30 to Acute Opioid Orders: The proportion among narcotic
orders placed for acute pain episodes where the morphine equivalent daily
dose exceeded 30
• Percentile Auth Provider Rate MEQD GT 30 to Acute Opioid Orders:
Percentile score of the provider compared with peers with lower percentiles
(in red shades) associated with higher MEQD >30 and higher percentiles (in
green shades) associated with lower MEQD>30
• Rate MEQD GT 80 to opioid orders: Proportion of all narcotic orders where
the morphine equivalent daily dose exceeded 80
• Percentile Auth Provider Rate MEQD GT 80 to Opioid Orders: Percentile
score of the provider compared with peers with lower percentiles (in red
shades) associated with higher MEQD>80 and higher percentiles (in green
shades) associated with lower MEQD>80
10
11
Two Opioid Key Performance
Indicators for 2018
1) Morphine Equivalent daily dose limit for acute pain
prescriptions
– Numerator: Number of total outpatient mode prescriptions where
morphine equivalent dose per day greater than 30
– Denominator: Acute outpatient opioid orders (in patients with no prior
opiate prescriptions in the last 100 days)
2) Opiate Burden
– Numerator: Total opiate burden (morphine equivalents)
– Denominator: Total unique patients with one or more selected
encounters
11
12
Region Baseline and Targets:
• Two metrics:
1. Rate MEDD > 30 to Acute Opioid Orders
2. Opioid Burden Rate
• 2018 targets are 90% of 2017 baseline
12
13
Users Guide to Opiate Data Cube
13
14
Performance on Metrics by Region
14
15
MEDD>30 for Acute Prescriptions
15
16
Opioid Burden
16
(Total MEDD for opioid prescriptions to number of unique patients)
17
Rate Morphine Equivalents >30
(Opioid naïve patients, Ohio law
8.31.17)
17
18
Opioid Burden: Total Morphine
equivalent divided by patient
18
19
2018 Opioid Prescribing Metrics by
Month
19
*Red box = Two strategic initiatives
20
Provider Dashboard
(Over time with drill down to orders)
20
21
Provider-level Graphs
21
22
Medication Assisted Therapy (MAT)
22
23
Prevention: Provider-Facing Tools
to Reduce Opioid Prescribing
23
24
State Level Responses to the
Opioid Crisis
• August 31, 2017 (Ohio)
– “7/5/30”
• No more than 7 days of opioids can be prescribed for adults and 5 days of opioids can be
prescribed for minors & only after the written consent of parent/guardian
• The total MEDD of a prescription for acute pain cannot exceed 30
• Rules apply to the first opioid analgesic prescription for the treatment of an episode of acute
pain
• November 15, 2017 (Kentucky)
– Limit of 3 day supply on C-II for acute pain
• December 29, 2017 (Ohio)
– Require diagnosis association on all opioid prescriptions
– Require indication of days supply on all controlled substance and gabapentin
prescriptions
• June 1, 2018 (Ohio)
– Require diagnosis association on all controlled substances
24
25
Preference List Customization
Creation via import of opioid specific facility preference list
containing fully configured compliant orders for selected common
acute opioids.
25
26
Preference List Customization
Providers can use either fully configured orders or select a less
configured choice.
26
27
In-Line MEDD Calculation
• All opioid containing medications possess a visual indicator of the calculated
MEDD within the order composer
• Dynamically calculates based on order dose and frequency
– Does not calculate
with free-text sigs
27
28
MEDD Calculation
Calculated Morphine Equivalent Daily Dose (MEDD)
• Have added a hyperlink in the upper right corner of our
Prescription Monitoring navigator section
• Cannot calculate with free-text sigs
28
29
Day Supply Limitations
Day Supply Designation
• Hard Stop
• Number of days will be dynamically
appended to sig when using discrete sigs
(default)
• Issue: Long-term meds, PRN frequency,
dispense qty, and duration may not always
align
29
30
30
Order Validation Alerts: MEDD
Threshold of 80 MEDD
c
c
Threshold of 30 MEDD
Order Validations fire at the END of the ordering workflow
• Content ONLY – provides stage directions for what is
suggested but no follow up actions via the popup
• If providers select “Accept” the order is placed
***Both MEDD and Days Supply can fire simultaneously
31
31
Order Validation: Days Supply
Ohio Adolescent: Days Supply = 5
Kentucky Patient: Days Supply = 3
c
Ohio Adult: Days Supply = 7
c
32
32
Unsigned Order BPA: MEDD
Threshold of 80 MEDD
c
c
Threshold of 30 MEDD
Unsigned Order BPA fires at the BEGINNING of the ordering workflow.
• It provides follow up actions such as removing orders, placing Naloxone prescription, and
links to document flowsheet values.
• If providers select “Accept” the order is NOT placed
33
33
Unsigned Order BPA: Days Supply
Ohio Adolescent: Days Supply = 5
Kentucky Patient: Days Supply = 3
c
Ohio Adult: Days Supply = 7
c
34
Controlled Substance Monitoring
SmartForm, Phase I
As a result of 2015 Ohio legislation, a SmartForm was implemented into Mercy’s EHR for quick
documentation that could be utilized for reports.
• Provided attestation button to document within the EMR that the PDMP report was reviewed -
a compliance requirement from PDMP
• Any documentation within the form could be pulled into the providers note via a SmartPhrase
• Provided a hyperlink out to the PDMP website to perform a manual query
34
35
Controlled Substance Monitoring
SmartForm Phase II: New Limits on
Prescription Opioids
• Effective August 31, 2017, Ohio passed new limits for prescribing opioids for acute pain
• SmartForm versatility allows additional documentation to the SmartForm already in use
• With provider guidance, the SmartForm was expanded to accommodate documenting the
new limits and exceptions on prescription opioids
35
36
Controlled Substance Monitoring
SmartForm Phase III: Appriss
Integration
1
2
3
4
5
6
1. Link to Integration and
SmartForm.
2. Prior recorded SmartForm
Values
3. MEDD Equivalent Daily
Dose Calculations
4. Urine Drug Screenings
5. Scanned Med Contracts
6. Flags related to FYI flags
7. Links to PDMP websites
7
36
37
Phase III: NarxCare Report
• The NarxCare report is the report
display from Appriss
– Takes the raw controlled
substance data received
from multiple state
pharmacies and creates
scores and graphs
– Providers can quickly tell
the patient's history with
opioids and likelihood for
abuse
37
38
Issue: Free Text Sigs
Solution: Require discrete sig for opioids
• Over 104,600 free text
orders in early 2 years
window
• Most can be
accommodated directly
with discrete frequency
38
39
Issue: Free Text Sigs
Solution: Dose and Frequency Range
Education
• Many providers were
unaware of ability to use
range doses on outpatient
prescriptions
• Created four new discrete
frequency range choices:
– Q3-4H PRN
– Q4-6H PRN
– Q6-8H PRN
– Q8-12H PRN
39
40
• Add additional information to
patient sig
– 140 total sig character limit
• Note to pharmacy
– 300 hard character limit
Issue: Free Text Sigs
Solution: Education of Conveying
Information to Pharmacy
40
41
Concomitant Benzodiazepine &
Opioid Prescribing Highest Risk for
Overdose and Death
• Actionable CDS for
Concomitant
Benzodiazepine & Opioid
Prescribing
41
42
Concomitant Benzo/Opioids:
Analytics
42
43
Pain Agreement Status in Order
Composer
• Per CMS, starting in 2019, Pain Agreements will expire for an existing
controlled substance after 6 months.
• When ordering a controlled substance, the pain agreement status will
display in the order composer.
43
Release Note 635758
44
Opioid Speed Buttons—Default of 3
Days
44
45
PDMP Review Information Available
on Radar Dashboards
• Help track individual & organization compliance with state regulations from
prescription drug monitoring programs.
• Show % of opioid prescriptions where providers didn't review PDMP
information in the same encounter.
Release Note 643819
45
46
• Contact information: Mabinstock@mercy.com
• Please complete online session evaluation
Questions
47
Appendix
47
48
• Gary Grazak, Integration Engineer
• Jedediah Tuten, Director Pharmacy,
Acute Operations
• Nicholas Waggamon, Application
Coordinator, Willow Pharmacist
• Karen Goda, Application
Coordinator, Ambulatory
• Anna Lendl, Application
Coordinator, Ambulatory
• Michael Temple, Manager
CarePATH, Research Informatics
• Marcus Hanna, Executive Director,
Emergency Services
Key Contributors
48
• Lisa Dubois, Application Coordinator,
Interfaces
• Matt Rasmussen, Integration Engineer II,
Cloverleaf
• Steve LeMaster, Application Coordinator,
Interfaces
• Wayne Bohenek, Vice President, Care
Transformation
• Brian Latham, Pharmacy Director, St.
Rita’s Medical Center
• Rob Quigley, Vice President, CarePATH
Ambulatory
• Kelley Recker, Vice President, CarePATH
Operations
49
Our Numbers:
49
50
Morphine Equivalents, cont.
Different opioids have different potencies, or Morphine
Equivalents (MEs):
Hydrocodone = 1
Oxycodone = 1.5
50
51
Opioid Prescribing – The Problem
• There were 500,000 prescriptions a year out of CarePATH for
Vicodin 5-500/Norco 5-325
– Each pill contains 5mg hydrocodone = 5.0 MEs
– Typical every 4 hour dosing (6 pills/day) = 30 MEs/day
• There were 500,000 prescriptions a year out of CarePATH for
Percocet 5 – 325
– Each pill contains 5mg oxycodone = 7.5 MEs
– Typical every 4 hour dosing (6 pills/day) = 45 MEs/day
These morphine equivalents were over the Ohio limit!
51
52
Analytics Methodologies - Build
• Key software used:
– SQL server 2014
– SQL Server Analysis Services 2016 RTM 1200
– Visual Studio DTS 2015
– Power BI v Oct 2015
– Excel 2016 with Power Pivot
– SSIS
• Data sources used:
– Epic (Clarity)
– Explorys (IBM)
– Kyruus
– ACO Payors
52
53
Acute Opioid
• Working definition: No preceding opioid order in a 100
day window prior to the incident opioid order.
– Measures not affected by erroneous or absent associated
diagnoses or problem list entries.
• All of these metrics were built in a manner that is not
dependent on the use of an Epic registry.
– They were set up in a way that was not dependent on, but
could leverage Epic’s method of calculating and storing
maximum morphine equivalent daily dose (MEDD)
53
54
Diagnosis Requirement Association
Order Validation
• Order Validation Point
– Hard stop for diagnosis association
– Lacks customization options
54
55
• Order validation lacks ability to report on firing rate trends and ability to measure success
of learned behavior
– Less alerts over time by way of providers associating upstream
• Created a robust tabular cube to monitor opioid prescribing overall allowed us to identify
an issue where some opioid orders were not requiring diagnosis
– Data sharing moved Kentucky market to elect to participate in requirement
Diagnosis Requirement Association
Order Validation
55
56
Actionable Alert: Special Consent (Ohio
HB 314)
• Clinical Decision
Support (CDS) for
special consent form
for opioid prescribing
to minors
56
57
Strategies to Decrease
Emergency Department Opioid
Use
57
58
0%
10%
20%
30%
Jan…
Mar…
Ma…
Jul '14
Sep…
Nov…
Jan…
Mar…
Ma…
Jul '15
Sep…
Nov…
Jan…
Mar…
Ma…
Jul '16
Sep…
Nov…
Jan…
Mar…
Ma…
Jul '17
Sep…
Nov…
Jan…
Mar…
Ma…
Jul '18
Emergency Department
Outpatient Opioid Prescription
Trend…
“Opioid-Free” Emergency Department
Provider Outreach
58
59
“Opioid-Free” Emergency Department:
Community and Patient Outreach
59
60
“Opioid-Free” Emergency Department
60
0%
5%
10%
15%
20%
25%
30%
Jan '14
Mar '14
May '14
Jul '14
Sep '14
Nov '14
Jan '15
Mar '15
May '15
Jul '15
Sep '15
Nov '15
Jan '16
Mar '16
May '16
Jul '16
Sep '16
Nov '16
Jan '17
Mar '17
May '17
Jul '17
Sep '17
Nov '17
Jan '18
Mar '18
May '18
Jul '18
Emergency Department Outpatient Opioid
Prescription Trend
Rate of Opioid Prescriptions to All Prescriptions
Springfield Market
Year Patient
Visits
Volume
Reduction
2014 80,916 -
2015 77,945 3.8%
2016 71,696 8%
2017 65,976 8%
61
Strategies to Decrease
Inpatient Opioid Use
61
62
Reducing Opioids in the Inpatient
Setting
• As part of an organizational focus on decreasing overall numbers
of opioid prescriptions, the following inpatient opportunities were
identified:
– Presence of narcotic pain relief options on admission order sets not
typically associated with pain
– Lack of a collection of Alternatives to Opioids (ALTO) options in one
concise format for ease in ordering
62
63
Designing a Solution:
Removing Pain Medications from Select
Order Sets
• In reviewing all admission and focused order sets with IV and oral
pain medication, it was determined that over 30 were for treatment
of diagnoses not normally associated with pain
– Validated with informatics committees and received nearly
unanimous support for removal of these pain medications
– Standard biennial review cycle of all order sets still containing
pain medications will be assessed for clinical appropriateness
moving forward
63
64
Designing a Solution:
Creating a Pain Management Focused
Order Set
• To supplement the removal of pain medications from many admission
order sets and to provide a single location for opioid and non-opioid pain
treatment a Pain Management Focused Order Set was created
• Plan to increase number of ALTO options after initial use period and
evaluation in conjunction with system Pharmacy and Therapeutics
Committee decisions
64
65
Designing a Solution:
Creating a Pain Management Focused
Order Set
• Key Features:
– Non-customizable
– Set as a suggested order set for all admitted patients
– In addition to traditional acetaminophen and ibuprofen, added
additional ALTO options
– Provider has to navigate through non-opioid options to get to
opioid choices
65
66
Designing a Solution:
Creating a Pain Management Focused
Order Set
66
67
SBIRT:
Screening, Brief Intervention, and
Referral to Treatment
• Prior to SBIRT Implementation:
• Alcohol, drug, and depression screenings were inconsistent
• Alcohol and drug use screenings were outdated and not
linked to action
• Approach to screening was inconsistent with public health
approach
• There were no outcomes captured for patients’ drug or
alcohol use
67
68
Screenshots: Prescreens
*All patients are asked the prescreen questions at triage
68
69
SBIRT Screening Numbers (2016 - Q1
2018)
0
10000
20000
30000
40000
50000
60000
70000
80000
2016 2017 2018
Q1 Q2 Q3 Q4 YEARLY TOTAL
**The large, sustained spike in screenings beginning in Q2 2017 is due to the spread of SBIRT to several additional
sites at that time. Additional sites were added in Q1 2018, resulting in another large spike in screenings.
69
70
SBIRT Outcomes - ROI
0
1
2
3
4
5
6
7
8
9
10
Alcohol Use Illegal Drug Use
Number of Days (last 30 days)
Change in Substance Use for Mercy Patients 6 months post
SBIRT screening protocol (n=155)
Baseline
Follow-up
The chart illustrates change in number of days of substance use for Mercy patients with both a baseline and follow-
up interview (n=155). Mercy patients demonstrated statistically significant reductions in alcohol (p=.002) and illegal
drug use (p=.001). Data was collected from December 2015-August 2017
70
71
Focused Order Set
Clinical Opiate Withdrawal
Scale (COWS)
71
72
Background
• Over the past few years, Mercy Health faced the rising tide of the
opioid crisis
• Amid the increasing numbers of opioid-related overdoses and
deaths, a growing number of patients with an opioid use disorder
were presenting at Mercy inpatient facilities
• There were few evidence-based, standardized tools to help guide
their treatment, and any existing tools were not integrated into the
EMR
• Lack of ability to manage the symptoms of these patients while
admitted to our facilities complicated care
– Increased burden of care on staff
– Increased number of patients unable to complete necessary treatment
for co-morbidities
72
73
Order Set Key Features
• Presence of buprenorphine
– Required approval by Mercy Health Formulary Committee for use in
order sets
– Limited to 72 hour duration to make available to physicians without
special prescribing authority to exceed 72 hours
• Combination of clonidine as adjunctive medication in linked panel
with both buprenorphine and tramadol as treatment options
• Fixed dose strategy
– COWS score dictated frequency of reassessment and follow-up doses
– Avoided confusion of titrating various doses of medication
• Availability of medications for symptom management
73
74
Opiate Withdrawal Focused Order Set
74
75
Designing a Solution:
Clinical Opiate Withdrawal Scale (COWS)
• Nursing assessment that
evaluates 11 signs/symptoms
• Stratifies severity of opiate
withdrawal
• Flowsheet built to auto-
calculate score
75
76
Outpatient Opioid Prescribing
Data Summary
-12%
-16%
-30%
-15
-50%
-20%
Significant Reductions in Opioid Prescribing Metrics
The following opioid ordering behaviors were substantially reduced
between December 2018 and December 2018:
Total Opioid
Orders
Opioid Orders to
All Medication
Orders
Opioids Orders
w/ MEDD > 80
Opioids w/ Day
Supply > 7 days
Total MEDD
per pt
Opioids
Orders w/
MEDD >30
Acute Pain
76
77
Success Stories
77
